analiza aiqbench
LIBRA addresses a critical bottleneck in agentic workflows by dynamically constructing executable tool libraries that capture prerequisite relationships without requiring manual dependency annotations or task-specific training. By validating tool compositions in a sandbox before user queries, the framework significantly improves retrieval recall and ranking metrics across multiple benchmarks compared to standard independent retrievers. This advancement enables more robust multi-step reasoning for large language model agents operating over complex tool catalogs.
LIBRA: Programmatic Induction of Executable Tool Libraries
arXiv:2611.05445v1 Announce Type: cross Abstract: Large language model agents increasingly retrieve tools from large catalogs, yet most retrievers rank each tool independently by request similarity. For multi tool tasks, this recovers the goal matching tool while omitting prerequisite tools required by the workflow.We introduce LIBRA, a framework for Library Induction by Runtime Assembly that turns a flat tool catalog into an executable tool library. An LLM proposes per tool contracts of candidate invocations, reusable outputs and state changes, and success conditions; LIBRA then tests these compositions in a sandbox and retains the relations supported at runtime. Constructed before user queries, the library requires no dependency annotations, tool use trajectories, or task specific training, and supports both one-shot and agentic retrieval. On AppWorld, BFCL, and SciToolEval, LIBRA improves one-shot Recall and NDCG over the same dense retriever by 7.9 and 3.7 percentage points on average. Within ToolReAGt, it yields corresponding gains of 4.5 and 2.9 points after ten retrieval iterations.
Abstract:Large language model agents increasingly retrieve tools from large catalogs, yet most retrievers rank each tool independently by request similarity. For multi tool tasks, this recovers the goal matching tool while omitting prerequisite tools required by the this http URL introduce LIBRA, a framework for Library Induction by Runtime Assembly that turns a flat tool catalog into an executable tool library. An LLM proposes per tool contracts of candidate invocations, reusable outputs and state changes, and success conditions; LIBRA then tests these compositions in a sandbox and retains the relations supported at runtime. Constructed before user queries, the library requires no dependency annotations, tool use trajectories, or task specific training, and supports both one-shot and agentic retrieval. On AppWorld, BFCL, and SciToolEval, LIBRA improves one-shot Recall and NDCG over the same dense retriever by 7.9 and 3.7 percentage points on average. Within ToolReAGt, it yields corresponding gains of 4.5 and 2.9 points after ten retrieval iterations.
10 pages, 5 figures. Introduces LIBRA, a query-agnostic tool library built via execution-validated prerequisites in a sandbox (avg ARR review score: 3.5/5). Demonstrates consistent performance gains across three benchmarks while discussing sandbox dependencies and library construction trade-offs
Software Engineering (cs.SE); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
arXiv:2611.05445 [cs.SE]
(or arXiv:2611.05445v1 [cs.SE] for this version)
https://doi.org/10.48550/arXiv.2611.05445 arXiv-issued DOI via DataCite (pending registration)
Text integral sub licența deschisă a publicației · 217 cuvinte
Subiecte conexe
ActionEngine: From Reactive to Programmatic Web Agents via State Machine Memory
arXiv · 30 sept. 2026
Implementing Computational Law in Wolfram Language for the Governance of Artificial Intelligence
arXiv · 17 aug. 2026
Co-Scraper: query-aware DOM Pruning and Reusable Scraper Synthesis for Lightweight Web Data Extraction
arXiv · 16 iun. 2026
SCI-AGENT-EXAM: Benchmarking Agents on Scientific Research Tasks
arXiv · 26 oct. 2026